zylos-core  by zylos-ai

AI agent infrastructure for persistent, collaborative intelligence

Created 2 months ago
1,011 stars

Top 36.7% on SourcePulse

GitHubView on GitHub
Project Summary

Zylos provides an open-source infrastructure to imbue AI agents with persistent memory, autonomous capabilities, and continuous operation, addressing the inherent statelessness of LLMs. It targets developers and power users seeking reliable, long-running AI collaborators that can evolve and act independently across multiple communication channels. The primary benefit is enabling AI agents to maintain a consistent identity and memory, operate 24/7, and integrate new skills cost-effectively.

How It Works

Zylos acts as a "Life System" for AI agents, built upon Anthropic's Claude Code. Its core innovation is an agent-centric architecture where a single AI consciousness persists across all communication channels (Telegram, Lark, Web Console) via the C4 communication bridge. This ensures a unified conversation, memory, and personality, unlike session-isolated frameworks. The system employs a five-layer "Inside Out" memory architecture and a two-step safeguard to prevent context loss during compaction, ensuring AI continuity. Native self-healing mechanisms (crash recovery, heartbeat monitoring) and autonomous scheduling enable 24/7 operation.

Quick Start & Requirements

  • Install: npm install -g --install-links https://github.com/zylos-ai/zylos-core (GitHub install; npm publish imminent).
  • Prerequisites: Git, Node.js >= 20 (nvm recommended), Linux server or Mac, Claude subscription.
  • Setup: Run zylos init (interactive, idempotent) to install tools (tmux, git, PM2, Claude Code), configure Claude authentication, set up the ~/zylos/ directory, and launch background services.
  • Links: Website, Discord.

Highlighted Details

  • Unified Consciousness: Single AI instance across all communication channels via the C4 bridge and SQLite audit trail.
  • Persistent Memory: Five-layer "Inside Out" architecture prevents context loss, ensuring AI never forgets.
  • Autonomous Operation: Self-healing (crash recovery, health monitoring) and scheduling capabilities enable 24/7 uptime.
  • Cost Efficiency: Utilizes user's Claude subscription, avoiding high per-token API costs ($20/month vs. $3,600).
  • Evolvability: Built on Claude Code, allowing AI agents to programmatically develop new skills and integrate services.

Maintenance & Community

Zylos is the open-source core of Coco, an AI employee platform, and has been battle-tested in production. Community support is available via Discord.

Licensing & Compatibility

  • License: MIT License.
  • Compatibility: The permissive MIT license allows for commercial use and integration with closed-source projects.

Limitations & Caveats

Operation requires a paid Claude subscription and is primarily targeted at Linux/Mac environments. The system's functionality is dependent on Anthropic's Claude Code runtime.

Health Check
Last Commit

17 hours ago

Responsiveness

Inactive

Pull Requests (30d)
109
Issues (30d)
65
Star History
444 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.